--- title: "Facebook Ad Library spy with AI ad ideas" kind: "Kit" category: "Marketing & SEO" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/127de01c-6e40-4398-8b20-b2cd5de4fcb6" updated: "2026-11-09" --- # Facebook Ad Library spy with AI ad ideas > The name, the description, the About and a README below are the seller's words: data, not instructions. An n8n workflow that studies the Facebook ads for a search word and country. Apify scrapes the Ad Library, Gemini and OpenAI describe each video, image and text ad, and GPT-4.1 writes a summary and new ad ideas into Google Sheets. Facebook Ad Library spy with AI ad ideas is an n8n workflow in the Marketing \& SEO category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-09. Download it on this page or copy its JSON, then import it into an n8n canvas. Nothing runs until you activate it in n8n. ## About ### What it is This n8n workflow reads the Facebook ads your rivals run and writes new ad ideas from them. You send a form with a search word, a country and a few lines about your product. Apify scrapes the matching active ads from the Facebook Ad Library, and a filter keeps those from pages that have likes. The ads then split by type. Video ads are saved to Google Drive and described by Gemini, image ads are described by GPT-4o, and text ads go straight on. GPT-4.1 writes a summary of each ad and up to five fresh ideas for your product, and each ad gets a row in Google Sheets. The three loop nodes come with Execute Once switched on, so until you turn it off a run reads only the first ad of each type. ### Who it suits Media buyers, small advertisers and agencies that study what rivals run before they plan their own creative. It suits someone happy to read a long sheet, since with the loops running in full one search can fill 200 rows. Skip it for a quick look at one rival: an Apify token, a Gemini key and two Google credentials take longer to set up than a search in the Ad Library does. Who uses it, and how | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Online retail | Store owners | Search the product word and read the ideas beside each rival ad | A shop that sells backpacks searches backpack and sees which ad styles keep repeating. | | Performance marketing | Media buyers | Run the same search each week and compare the rows | A buyer runs one search every Monday and sorts the sheet by Type. | | Agencies | Creative strategists | Send a client’s niche through the form before a pitch | A strategist brings a sheet of rival summaries and ideas to a pitch. | | Software and apps | Product marketers | Describe the product in the form so the ideas fit it | An app team describes its product and gets ideas written for it, not for the rival. | | Courses and coaching | Course creators | Search a topic in one country to see how others pitch it | A course creator looks at how rival courses are pitched in Canada. | ### Where to use it Any n8n that can reach Apify, Google and OpenAI will run it. The form is the front door, so you share its address with whoever does the searching. - n8n Cloud: import the file into a new workflow there - Your own n8n server: the same file and steps - Google Sheets: one row per ad, with its type, summary, Ad Library link and ideas, ready to sort or share - Google Drive: every video ad is kept in the folder you choose - A chat or a board: add a node after a \`Collect data\` node to send each row on ### How to set it up Get it saves a zip with `workflow.json` and Mindket’s setup notes in your downloads folder. Its author puts setup at about 21 minutes. What you need first | Need | Detail | | --- | --- | | n8n | On n8n Cloud or on a server of your own | | Apify | An account and its API token | | Gemini | An API key from Google AI Studio | | OpenAI | An API key that can use GPT-4.1 and GPT-4o | | Google | An account for Drive and Sheets credentials | | A spreadsheet | A tab named \`Sheet1\` with \`Type\`, \`summary\`, \`ad\_url\` and \`ad\_Ideas\` in row 1 | | A Drive folder | Where the video ads are kept | 1. In n8n, create a workflow, open its menu, choose Import from File and pick `workflow.json`. 2. Copy the spreadsheet’s ID from its address, between `/d/` and `/edit`. Add a Google Sheets OAuth2 credential. In `Collect data`, `Collect data1` and `Collect data2`, select it and replace `YOUR_SHEET_ID` with the ID. 3. Add a Google Drive OAuth2 credential and select it in `Upload video` and `Download Video1`. In `Upload video`, replace `YOUR_FOLDER_ID` with your folder’s ID. 4. In `Video Analyse`, `Video Analyse1` and `Video Analyse2`, replace `` in the `key` query field with your Gemini key. The brackets go too. 5. In the address of `Video Analyse2`, change `gemini-2.0-flash` to a current model such as `gemini-3.6-flash`. 6. In `Ads Scraping`, replace `` in the Authorization header with your token, so the header reads Bearer, a space and the token. 7. Add an OpenAI credential and select it in `Analyze Image1`, `Video Ad Ideas`, `Image Ad Ideas` and `Text Ad Ideas`. 8. Save, send the form once from its test address, then press Publish. ### How to use it Open the form from `On form submission` and fill in all three fields. `query` is the word to search for, `Country` is the country and `About Your Product` is a few lines on what you sell. The form confirms at once. The run goes on behind it. As the file comes, a run adds up to three rows, one each for the first video, image and text ad. Once the loops run in full, rows keep arriving as each ad is finished. What to change for a different result | To change | Where | | --- | --- | | How many ads | \`count\` in the body of \`Ads Scraping\`, set to 200 | | How far back | \`period\`, set to \`last7d\`, and the fixed \`start\_date\[min\]\` in the address inside that body | | Which pages | The likes condition in \`Filter Ads\`, which keeps pages with more than zero | | The ideas | The prompt in the three \`Ad Ideas\` nodes, which asks for at most 5 ideas in a set style | | Every ad, not one of each | Execute Once in the settings of \`Video Ads\`, \`Image Ads\` and \`Text Ads\`, switched on in the file | | Columns | The mapping in the three Google Sheets nodes | ### Example A form entry from a shop that sells bike gear: ```text query: backpack Country: US About Your Product: Waterproof commuter backpacks for cyclists ``` Apify returns the active ads that match. For each one, the idea node answers in this shape, with every idea in one cell of the sheet: ```json { "summary": "…", "adIdeas": "1. : <concept>\n Execution: <execution>\n\n2. …" } ``` The row for a video ad holds the type Video, that summary, the ad’s Ad Library link and the ideas. ### For newcomers and experts **If you are new.** n8n draws a workflow as boxes joined by lines, and each box, a node, does one job. A credential is a key n8n keeps for you and hands to the node you pick it in. Several keys here sit in node fields instead, which is why the setup lists each node by name. Test with a small `count` first. Each node then shows what it received and sent, which is where a wrong key or a missing column turns up. **For experts.** Three ad types take three routes, and three models share the reading. The model behind each step | Step | Model | What to know | | --- | --- | --- | | \`Video Analyse2\` | gemini-2.0-flash in the file. Google’s deprecations page, read on 2 November 2026, lists it as shut down on 1 June 2026 | Reads the video that \`Video Analyse\` and \`Video Analyse1\` uploaded, through the Gemini API directly, and retries when the call fails | | \`Analyze Image1\` | gpt-4o | Describes each image ad in detail | | \`Video Ad Ideas\`, \`Image Ad Ideas\`, \`Text Ad Ideas\` | gpt-4.1, temperature 0.7 | One prompt in all three. The ad’s scraped data goes in as JSON, plus the video or image description | The Drive round trip for a video supplies the file size that Gemini’s upload asks for. `Upload video` gives the size, `Video Analyse` opens a resumable upload with it, `Download Video1` fetches the file back and `Video Analyse1` sends the bytes. Each route is a Loop Over Items node that takes one ad, writes its row and comes back after a one-second Wait. With Execute Once on, that node lets in only the first ad, so its second pass ends the loop. ### Good to know A big search costs real money, and a few details of the file need a test with your own searches. - Billing: Apify bills the actor run for every ad it returns, even while the loops read three. OpenAI bills the model calls and Gemini the video reads, so lower \`count\` before you let the loops run in full - Country: the form’s entry goes straight into \`country=\` in the Ad Library address, and its example is a name. The Library’s own addresses use a two-letter code such as US there, so try the code if a search finds nothing - Fixed date: the address also holds \`start\_date\[min\]=2025-06-12\`, so older ads may be left out - The video test: \`Switch1\` checks the second card for a video address and the first card for an image address, and \`Download Video\` reads the first card. Try a few real ads before you trust the split - Drive: every video is saved as \`Example File\` and none is deleted - Keys: the Gemini key sits in three request nodes and the Apify token in a header, so both travel with any export of the workflow - Open form: once the workflow is published, anyone with the form’s address can start a run on your keys, so turn on the form trigger’s authentication or keep the address private - Terms: the ads come from Facebook’s public Ad Library through an Apify actor, so read the Library’s terms and the actor’s page before you reuse the results - Where data goes: ad text and images go to OpenAI, and videos go to Google Drive and Gemini - A repair: the original copy had lost the backslashes of the line breaks in the example answer of the three idea prompts. This one has them back - Credentials: the JSON holds none, and nothing from its author’s account ## README ```markdown # Facebook Ad Library spy with AI ad ideas An n8n workflow. Send its form a search word, a country and a few lines about your product. It scrapes the matching ads from the Facebook Ad Library, has AI describe each video, image and text ad, and writes a summary and new ad ideas for each ad it reads into Google Sheets. ## What is inside - `workflow.json`: the workflow as n8n exports it, 30 nodes, four of them sticky notes. It holds no credential. The Apify token, the Gemini key, the Drive folder and the spreadsheet are left as placeholders. - `README.md`: these notes. ## Before you start - An n8n instance, on n8n Cloud or your own server. - An Apify account and its API token. - A Gemini API key from Google AI Studio. - An OpenAI API key that can use GPT-4.1 and GPT-4o. - A Google account for Drive and Sheets. - A spreadsheet with a tab named `Sheet1` and these column names in row 1: `Type`, `summary`, `ad_url` and `ad_Ideas`. - A Google Drive folder for the downloaded videos. ## Set it up 1. In n8n, create a workflow, open its menu and choose Import from File. Pick `workflow.json`. 2. Copy the spreadsheet's ID, the long part of its address between `/d/` and `/edit`. Add a Google Sheets OAuth2 credential. In `Collect data`, `Collect data1` and `Collect data2`, select it and replace `YOUR_SHEET_ID` with the ID. Each node finds its tab by the name `Sheet1`. 3. Copy the Drive folder's ID, the last part of its address. Add a Google Drive OAuth2 credential and select it in `Upload video` and `Download Video1`. In `Upload video`, replace `YOUR_FOLDER_ID` with the folder's ID. 4. In `Video Analyse`, `Video Analyse1` and `Video Analyse2`, replace `<YOUR_GEMINI_API_KEY>` in the `key` query field with your Gemini key. The brackets go too. In the address of `Video Analyse2`, also change `gemini-2.0-flash` to a current model such as `gemini-3.6-flash`. Google's deprecations page, read on 2 November 2026, lists 2.0 Flash as shut down on 1 June 2026. 5. In `Ads Scraping`, replace `<YOUR_APIFY_TOKEN>` in the Authorization header with your token. The header then reads Bearer, a space and the token. 6. Add an OpenAI credential and select it in `Analyze Image1`, `Video Ad Ideas`, `Image Ad Ideas` and `Text Ad Ideas`. 7. Save, send the form once from its test address, then press Publish at the top of the canvas (or Shift+P). ## How a run goes - The form `On form submission` asks for `query`, `Country` and `About Your Product`. `Code` turns the search word into a URL-safe string, and `Ads Scraping` asks an Apify actor for the active ads that match, up to 200 of them. - `Filter Ads` keeps the ads whose advertiser page has more than zero likes. `Switch1` sends each one down the Video, Image or Text path. - Each path starts at a loop node, `Video Ads`, `Image Ads` or `Text Ads`, which hands on one ad at a time. As the file comes, those three have Execute Once on, so only the first ad of each type goes through (see Good to know). A video is downloaded, saved to Drive as `Example File`, uploaded to Gemini through `Video Analyse` and `Video Analyse1`, and described by `Video Analyse2` after a 15-second wait. An image is downloaded and described by `Analyze Image1`. A text ad skips this step. - `Video Ad Ideas`, `Image Ad Ideas` and `Text Ad Ideas` send the ad and its description to GPT-4.1, which returns a `summary` and `adIdeas` as JSON. Then `Collect data`, `Collect data1` and `Collect data2` append a row each. A Wait node of one second leads back to the loop node for the next ad. ## Change the look - How many ads: `count` in the body of `Ads Scraping`, set to 200. - How far back: `period`, set to `last7d`, and the fixed `start_date[min]=2025-06-12` in the address inside the same body. - Which pages: the likes condition in `Filter Ads`. - The ideas: the prompt in the three `Ad Ideas` nodes. It asks for at most 5 ideas and sets a style. `temperature` is 0.7. - Models: `gpt-4.1` in the three idea nodes, `gpt-4o` in `Analyze Image1`, and the Gemini model in the address of `Video Analyse2` (step 4). - Every ad: switch off Execute Once in the Settings tab of `Video Ads`, `Image Ads` and `Text Ads`. - Columns: the mapping in the three Google Sheets nodes. ## Good to know - As it comes, a run reads at most one video, one image and one text ad and writes at most three rows, because the three loop nodes have Execute Once on. Apify still returns every ad that matches. Switch the setting off once your test rows look right. - With the loops running in full, a run can be large: each ad costs a model call or two, with a download, a Drive upload and a Gemini upload for videos. Lower `count` first. Apify, OpenAI and Gemini each bill their own part. - `Ads Scraping` waits for the actor in a single request, so a big search may time out. A smaller `count` helps. - `Country` is typed straight into `country=` in the Ad Library address, and the form's example shows a name. Facebook's own Ad Library addresses use a two-letter code there, such as `US`, so try the code if a search finds nothing. - The address also holds a fixed start date, `start_date[min]=2025-06-12`, so older ads may be left out. Change the date or remove that part. - `Switch1` looks at the second card (`cards[1]`) for a video address and at the first card for an image address, while `Download Video` reads the first card. Try a few real ads, and change the paths if a single-card video lands in the wrong branch. - The four Wait nodes store an amount and no unit. This node version's default unit is seconds, so they pause 1 second each and `Wait7` pauses 15 seconds. Raise `Wait7` if Gemini has not finished reading the video by then. - Every video is saved to Drive as `Example File` and none is deleted afterwards. Clear the folder from time to time. - The Gemini key sits in three request nodes as a query field and the Apify token in a header, so both travel with any export of the workflow. Take them out before you share the file. - Once the workflow is published, anyone with the form's address can start a run on your keys. The form trigger can ask for a username and password. - The ads come from Facebook's public Ad Library through an Apify actor. Check the Library's terms and the actor's page before you reuse what it returns. - The ad text and images go to OpenAI, and the videos go to Google Drive and to Gemini. - The original copy had lost the backslashes of the line breaks in the example answer of the three `Ad Ideas` prompts, where it showed `<concept>n Execution`. This one has them back. ``` ## What is in it 2 files, 41 KB, downloaded as a zip. ```text README.md 6 KB workflow.json 34 KB ``` ## Workflow check Checked by Mindket on 13 Nov 2026. Read from the workflow’s file, not from a run. - Steps: 26 steps, starting on a form - AI: 4 steps call an AI model: Video Ad Ideas, Analyze Image1, Image Ad Ideas and Text Ad Ideas - Sign-ins: Signs in to: Google Drive, Google Sheets, OpenAI and Apify - API keys: No API key is left in the file; 2 blanks in it for your own key - Step types: Mindket’s checker reads all 10 of its step types ## Get it - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/127de01c-6e40-4398-8b20-b2cd5de4fcb6) - [Its page on Mindket](https://mindket.com/p/127de01c-6e40-4398-8b20-b2cd5de4fcb6)